𝗗𝗮𝘆-𝟭𝟵𝟬 Computer Vision Learning 𝗔𝗱𝘃𝗣𝗖: Transferable Adversarial Perturbations on 3D Point Clouds by King Abdullah University of Science and Technology (KAUST), Thuwal, Saudi Arabia Follow me for similar post : 🇮🇳 Ashish Patel Interesting Facts : 🔸 This is a paper in ECCV2020 with over 15 citations. 🔸 It outperforms PointNet, PointNet++ etc. ------------------------------------------------------------------- 𝗔𝗺𝗮𝘇𝗶𝗻𝗴 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵 : https://lnkd.in/gfnM82e Code : https://lnkd.in/g5NVFkd ------------------------------------------------------------------- 𝗜𝗠𝗣𝗢𝗥𝗧𝗔𝗡𝗖𝗘 🔸 Deep neural networks are vulnerable to adversarial attacks, in which imperceptible perturbations to their input lead to erroneous network predictions. This phenomenon has been extensively studied in the image domain, and has only recently been extended to 3D point clouds. 🔸In this work, They present novel data-driven adversarial attacks against 3D point cloud networks. They aim to address the following problems in current 3D point cloud adversarial attacks: they do not transfer well between different networks, and they are easy to defend against via simple statistical methods. 🔸To this extent, They develop a new point cloud attack (dubbed AdvPC) that exploits the input data distribution by adding an adversarial loss, after Auto-Encoder reconstruction, to the objective it optimizes. AdvPC leads to perturbations that are resilient against current defenses, while remaining highly transferable compared to state-of-the-art attacks. 🔸They test AdvPC using four popular point cloud networks: PointNet, PointNet++ (MSG and SSG), and DGCNN. #computervision #artificialintelligence #data